Emerging latency-sensitive and data-intensive applications-such as autonomous unmanned systems, AR/VR, and immersive digital services-demand intelligent orchestration of interdependent services across heterogeneous and geographically distributed cloud–edge infrastructures. This paper presents a novel AI-driven, QoS-aware orchestration framework designed for the dynamic management of such services within the cloud continuum. The framework introduces a unified modeling approach that captures both service inter-dependencies and shared resource constraints, facilitating SLA- and ALA-driven orchestration. A proactive placement algorithm, enhanced with predictive mechanisms such as workload forecasting and latency trend analysis, ensures compliance with stringent QoS requirements while adapting to the volatility of mobile and heterogeneous environments. The proposed online orchestration system dynamically monitors environmental and workload conditions, enabling real-time service (re)placement to mitigate QoS degradation and optimize resource usage. A proof-of-concept implementation, validated through a wildfire surveillance use case with autonomous drones, demonstrates significant improvements in service availability and responsiveness, advancing the state-of-the-art in continuum computing.